Epidemiologic studies of antenatal caffeine consumption and adverse pregnancy outcomes have produced conflicting results. Most prior studies relied on self-reported caffeine consumption as a measure of caffeine exposure. However, caffeine metabolism and clearance greatly affect internal dose and, subsequently, fetal dose and consumption data may not provide a valid assessment of caffeine exposure. Because only 2.0% of caffeine is excreted as such in urine, there is considerable opportunity for error in estimating caffeine exposure using urinary caffeine levels. Paraxanthine, theophylline, and theobromine are primary metabolites of caffeine in humans and readily detected in body fluids. These metabolites may provide more reliable estimates of the biologically effective internal dose of caffeine. The specific aims of this application are: 1) to examine associations between intra-uterine growth retardation, preterm delivery, and birthweight with fetal caffeine exposure as estimated by cord blood serum paraxanthine; 2) to examine associations between serum paraxanthine, urinary caffeine, other metabolites, and self-reported caffeine intake; 3) to develop statistical models to determine the most precise predictive factors of caffeine exposure. We will link the analysis of serum caffeine metabolites with data previously collected on a large cohort of pregnant women (n=1775). Trained research assistants administered a baseline interview that included detailed questions on caffeine and decaffeinated beverage consumption, demographics, pregnancy history, medical history, tobacco and alcohol use, physical activity, use of nutritional supplements, and other reproductive risk factors. All women were asked to provide a urine sample at a baseline interview and cord blood was routinely collected and stored. In addition, women were randomly assigned to provide urine samples at 20, 28 or 36 weeks gestation. Medical records were abstracted to obtain information on obstetrical outcomes. The proposed work would conduct detailed biomarker analyses on samples collected from this cohort to provide a more complete understanding of effects of caffeine exposure on perinatal outcomes from a more accurate estimate of fetal exposure.